STEADY STATE AND TRANSITION STATE MULTI-VARIABLE CONTROL METHOD OF A TURBOFAN ENGINE BASED ON AN ACTIVE DISTURBANCE REJECTION CONTROL THEORY (ADRC)
A steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC) belongs to the technical field of aero-engine control. Firstly, multiple groups of manipulated variables and controlled variables of a turbofan engine are preliminarily selected, and then the manipulated variables and controlled variables with high correlation are further determined by a correlation analysis method. Each group of control instructions is planned as a tracking trajectory by using a tracking differentiator, and is used as the input of a nonlinear proportional-differential controller together with the current state estimated by the extended state observer to calculate the manipulated variables. This achieves the objective that a turbofan engine requires multi-input and multi-output to operate simultaneously according to a predetermined trajectory and meet the control requirements, and has the advantages of less difficult controller design method, fewer parameters to be tuned.
The present invention relates to a steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC), in particular to establishing a multi-variable control framework of an aero-engine to optimize the control effect of the turbofan engine in the process of operating at idle and above speeds based on the ADRC, belonging to the technical field of aero-engine control.
BACKGROUNDThe present invention replies on the background of the control technology of a certain type of twin-spool turbofan engine. At present, the relevant technology of aero-engines in China mostly improves the efficiency of an aero-engine from the perspectives of material and structure, but seldom makes improvement and optimization in the field of control. Although the improvement in material and structure can improve the efficiency of a turbofan engine, the optimization in the field of control can give better play to the potential of the current aero-engine and extend the service life of the turbofan engine. In the early design stage of the aero-engine, a controller thereof is realized in the form of a mechanical device and a hydraulic structure; although the stability is good, only a simple control strategy can be realized and no complex control algorithm can be applied. With the improvement of digital electronic technology, a full authority digital electronic controller (FADEC) is gradually applied to the specific implementation of a turbofan engine control system, and has many advantages such as that the control strategy is easy to modify and complex algorithms are able to be realized.
The control of the turbofan engine can be divided into several stages, such as start/stop control, steady state control, accelerate/decelerate control, and augmentation control according to the different stages which the turbofan engine is in. The present invention does not relate to a start/stop control part.
Steady state control refers to the control process of maintaining the engine speed unchanged at any point in the process from idle speed to maximum speed of the turbofan engine. A PID control algorithm is mostly adopted in the steady state control of a traditional turbofan engine, which requires to obtain a linear model for small deviations of the turbofan engine at each steady state point, tune PID parameters for all linear models to obtain the increment of turbofan engine fuel, and act together with steady state fuel as actual fuel input variables of the turbofan engine. This method is simple in principle, but involves many aspects such as linear model identification, PID parameter tuning and gain scheduling, and the design process is quite complex.
Accelerate/decelerate control refers to the control process of a gas turbine changing rapidly between the idle speed state and the rated speed state. The aerodynamic and thermodynamic characteristics of the turbofan engine change greatly in this process, so the above-mentioned steady state controller is not competent for the work in an accelerate/decelerate process, and an accelerate/decelerate schedule is usually used in an actual process to limit the operation of the turbofan engine to a reasonable range. Once the turbofan engine is in the accelerate/decelerate schedule, the controller of the turbofan engine will give control authority to the accelerate/decelerate schedule, and the control effect will be determined by the preset accelerate/decelerate schedule.
In addition, with the increasing requirement of flexibility and working efficiency, the multi-variable control of the turbofan engine is becoming a breakthrough direction. In particular, the introduction of an afterburner in the turbofan engine makes it impossible to maintain the stable state of the turbofan engine only in the manner of tuning combustion chamber fuel. At present, the relevant actions of the turbofan engine after the afterburner is started are also schedule-based, and the control effect depends on the setting of the schedule in early stage. The multi-variable control of the turbofan engine can provide a new optimization space for aero-engine control, is beneficial to improving the comprehensive performance of the turbofan engine, and gives full play to the potential of the engine in operating process.
To sum up, the current turbofan engine control technology has many shortcomings. Firstly, the design of a turbofan engine controller is mainly focused on the design of the steady state controller, and the working boundary of the turbofan engine is limited by different schedules during accelerate/decelerate and augmentation period, so the design of the controller can only ensure the steady state performance but cannot change the transition state performance thereof; secondly, there is a problem of control authority exchange between the steady state controller and the accelerate/decelerate schedule in the operating process of the turbofan engine, resulting in that the control strategy of the controller in steady state is not unified with that in accelerate/decelerate state, so both the steady state controller and a reasonable transition state schedule as well as an action schedule of an augmentation process actuator needs to be designed; in addition, the current turbofan engine control strategy does not meet the requirements of multi-variable control, and the structure of the current control system needs to be modified a lot to meet the new requirements such as decoupling; finally, the existing turbofan engine control system needs to tune multiple groups of parameters and a large number of schedule curves, and the design process of the controller is cumbersome and inefficient.
Therefore, in order to overcome the shortcomings of the current aero-engine control system design, the present invention designs a multi-variable control system of a turbofan engine based on the ADRC. The multi-variable control system of the present invention not only treats the steady state and transition state of the turbofan engine uniformly as the same condition, but also avoids the uncertainty introduced by switching between different methods. In addition, the present invention uses the method of total disturbance estimation to uniformly observe the internal state change and external disturbance influence of the turbofan engine, and implements rejection offset, which has better robustness than the traditional method. The present invention is also applicable to gas turbines and other devices with the same functional structure or similar operating characteristic lines.
SUMMARYAiming at multiple problems of the existing method such as that the design process of the turbofan engine control system is complex and multi-variable control is difficult to realize depending on the schedule, the present invention provides a steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC).
The technical solution of the present invention is:
A steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC), comprising the following steps:
S1. selecting two or more groups of controlled variables and determining the control parameter requirements of each controlled variable based on multi-variable control objectives, and then determining a manipulated variable corresponding to each controlled variable in combination with a mechanism analysis method and a correlation analysis method;
the steps of determining the controlled variables and the manipulated variables are as follows:
S1.1 analyzing the control requirements of a turbofan engine, and preliminarily determining the manipulated variables U=[u1, . . . , ui, . . . , un]T and the controlled variables Y=[y1, . . . , yi, . . . , yn]T according to the mechanism of the turbofan engine, wherein ui and yi are the variables of the same group in the ith loop;
S1.2 selecting the ith group of manipulated variables and controlled variables to perform analysis, maintaining other input variables [u1, . . . , ui−1, ui+1, . . . , un]T of the turbofan engine as reasonable constants, and setting the sequence of the ith group of manipulated variables to
ũi=[ũi,1, ũi,2, . . . , ũi,k, . . . , ũi,m]
where: ũi,k−1<ũi,k, ũi,k is within the normal operation range of the manipulated variable ui, and m means the number of data sampling points in this group;
S1.3 running a turbofan engine model to obtain output data
{tilde over (y)}i=[{tilde over (y)}i,1, . . . {tilde over (y)}i,m]
S1.4 normalizing the obtained input ũi and output {tilde over (y)}i of the turbofan engine as follows:
S1.5 using the correlation analysis method to calculate a correlation coefficient Rei as follows:
S1.6 determining whether the manipulated variable ui is used for controlling the controlled variable yi according to the obtained correlation coefficient Rei; if the correlation coefficient is satisfactory, this group of parameters is selected correctly; otherwise, if the correlation between the manipulated variable and the controlled variable is not significant, the manipulated variable needs to be replaced until the correlation coefficient is satisfactory;
S2. establishing a tracking differentiator (TD) module, a linear extended state observation module and a nonlinear PD module according to the basic principle of active disturbance rejection control, building a steady state and transition state controller, and reserving the time constant rt of the TD, wo of the linear extended state observer (LESO), and Kp and Kd of the nonlinear PD as the parameters to be tuned;
the steps of establishing the steady state and transition state controller are as follows:
S2.1 establishing a fastest control integrated function ƒhan(p,q,rt,h) of a second-order discrete system, and the expressions thereof are as follows:
where, rt is a parameter to be tuned and is used for tuning the length of the transition time for the output variables of the TD; the greater rt is, the shorter the transient time is; h is a simulation step size; sign(ƒx) is a sign function,
d, d0, ƒx, a0, a are internal variables introduced for the convenience of calculation, and ƒhan is the output of the function fhan(p,q,rt,h);
S2.2 building the TD module according to the established function ƒhan(p,q,rt,h), the input of the TD is a control instruction vi of the ith loop, the output is respectively a tracking trajectory gi and a trajectory derivative gi′, and the discrete-form updated expressions thereof are as follows:
gi=gi+h×gi′
gi′=gi′+h׃han(gi−vi, gi′, rt,i, h)
S2.3 establishing the LESO, using the input ui and output yi of the turbofan engine as the input of extended state observation, and observing the states ŷe,i and Sŷe,i′ of the current output variables and the total disturbance ŷtotal,i; the expressions thereof are shown as follows:
where, b0,i and wo,i are respectively parameters of the extended state observer (ESO), b0,i is a model characterization parameter and is related to an actual model, and wo,i is a bandwidth parameter of the ESO; ui and yi are the input of the ESO, Z is a state variable of the ESO, Ŷ is the output of the ESO, and three output variables are respectively the estimated value ŷe,i of the output yi of the turbofan engine, the changing trend ŷe,i′ of yi, and the total disturbance ŷtotal,i of the turbofan engine in this loop;
S2.4 establishing a nonlinear function fal(e,α,δ), and the expression thereof is shown as follows:
S2.5 establishing a nonlinear PD feedback controller shown as follows in accordance with the established nonlinear function:
uPD,i=Kp,iƒal(ei, αp,iδp,i)+Kd,iƒal(ei′, αd,i, δd,i)
S2.6 assembling and establishing the steady state and transition state controller of the turbofan engine in accordance with the structure of each module in the ADRC;
S3. selecting a group of manipulated variables and controlled variables as a tuned parameter control loop, maintaining the manipulated variables of the remaining loops unchanged or operate in accordance with a desired trajectory, and tuning the parameters of the steady state and transition state controller to basically meet the control requirements;
the steps of tuning the ith group of control parameters of the steady state and transition state controller are as follows:
S3.1 preliminarily selecting the parameters of the ESO according to the data ũi and {tilde over (y)}i of the turbofan engine to meet the condition of the following expression:
-
- S3.2 using the data ũi and {tilde over (y)}i of the turbofan engine as the input of the ESO, and tuning the parameter wo,i of the ESO to make the output ŷe,i of the ESO track {tilde over (y)}i correctly;
S3.3 inputting a reference instruction vi, and tuning a parameter rt,i to obtain a desired transition trajectory gi; when the trajectory transition time is too long, increasing the parameter rt,i; otherwise, decreasing rt,i;
S3.4 using ei=gi−ŷe,i and ei′=gi′−ŷe,i′ as the input of a nonlinear PD controller, and tuning Kp,i and Kd,i to reasonable values; the output thereof is uPD,i;
S3.5 calculating the output of the steady state and transition state controller as
S3.6 forming a closed loop by the controller and the turbofan engine, and then comprehensively tuning Kp,i, Kd,i, wo,i and b0,i to ensure that the controlled variable yi can track a given trajectory well;
S4. establishing an upper limit protection function funl(xi, xi,l, xi,dl, uj,l), designing a state/performance parameter protection controller of the turbofan engine based on this function, and limiting the output of the control loop to ensure that the controlled variables of the system meet the control requirements while the state parameters thereof do not exceed the limit;
the steps of establishing the state/performance parameter protection controller of the turbofan engine are as follows:
S4.1 determining the limiting value xi,l of a limit protection parameter and the action range xi,dl of a limit protection controller according to the control objectives of the turbofan engine;
S4.2 establishing the upper limit protection function ƒunl(xi, xi,l, xi,dl, uj,l),and the formula thereof is as follows:
where, xi is a turbofan engine parameter requiring limit protection, xi,l is the allowable maximum value of xi, xi,dl is the size of the operating range of the limit protection controller, i.e. the controller comes into action when xi>x,l−xi,dl, and uj,l means the maximum value that the limit protection controller can output, wherein j means that the controller output corresponding to the limit protection parameter is not related to the sequence of the manipulated variables of the turbofan engine;
S4.3 establishing a discrete-state dead band link based on the limit protection function, and the expression thereof is shown as follows:
where, ƒunout (k) means the output of the limit protection controller at the kth time point, xi,k, xi,k−1, xi,k−2 are respectively the numerical values of the turbofan engine parameter at the kth, (k-1)th and (k-2)th time points, and Δi and Δi ′ are respectively the sizes of a position dead band and a speed dead band;
S4.4 connecting the output of the limit protection controller to the control loop in the form of negative feedback, and tuning the maximum value uj,l that the limit protection controller can output to ensure that the limit protection controller can fulfill the role of over-limit protection;
S5. conducting a multi-variable controller test, and finely tuning each parameter to ensure the overall effect of the turbofan engine so as to enhance the adaptability of a multi-variable control system of the turbofan engine to the change of external environment.
The present invention has the following beneficial effects that: the steady state and transition state multi-variable control system of a turbofan engine of the design of the present invention can not only achieve the objective that a turbofan engine requires multi-input and multi-output to operate simultaneously according to a predetermined trajectory and meet the control requirements, but also has the advantages of less difficult controller design method, fewer parameters to be tuned, clearer physical significances and greatly improved system robustness compared with the traditional control. Therefore, the present invention provides a new and more effective control idea for the multi-variable control of the turbofan engine, which can establish a real-time limit protection controller on the basis of meeting the multi-variable control requirement, steady state control requirement, servo control requirement and disturbance rejection control performance requirement of the turbofan engine to ensure that the turbofan engine operates within a safe range at all time points and prevent the turbofan engine from dangers. At the same time, the estimation ability of the ADRC for unknown disturbances is fully used in this method to take the coupling among the loops in the multi-variable control of the turbofan engine as the total disturbance for estimation, so no special treatment for the coupling part is needed. This method can not only completely replace the traditional control strategy based on PID, but also cooperate with the traditional limit protection strategy with min-max as the core, with flexible and diverse implementation methods and simple logic structure. At the same time, this method is also applicable to the design of the control systems of gas turbines with a similar structure and internal combustion engines with a similar working principle, and the application range is wide.
The present invention is further described below in combination with the drawings. The present invention replies on the background of a nonlinear model of a certain type of twin-spool turbofan engine, and the control structure diagram is shown in
As shown in
S1. selecting two or more groups of controlled variables and determining the control parameter requirements of each controlled variable based on multi-variable control objectives, and then determining a manipulated variable corresponding to each controlled variable in combination with a mechanism analysis method and a correlation analysis method;
S2. establishing a tracking differentiator (TD) module, a linear extended state observation module and a nonlinear PD module according to the basic principle of active disturbance rejection control, building a steady state and transition state controller, and reserving the time constant rt of the TD, wo of the linear extended state observer (LESO), and Kp and Kd of the nonlinear PD as the parameters to be tuned;
S3. selecting a group of manipulated variables and controlled variables as a tuned parameter control loop, maintaining the manipulated variables of the remaining loops unchanged or operate in accordance with a desired trajectory, and tuning the parameters of the steady state and transition state controller to basically meet the control requirements;
S4. establishing an upper limit protection function ƒunl(xi, xi,l, xi,dl, uj,l), designing a state/performance parameter protection controller of the turbofan engine based on this function, and limiting the output of the control loop to ensure that the controlled variables of the system meet the control requirements while the state parameters thereof do not exceed the limit;
S5. conducting a multi-variable controller test, and finely tuning each parameter to ensure the overall effect of the turbofan engine so as to enhance the adaptability of a multi-variable control system of the turbofan engine to the change of external environment.
As shown in
S1. analyzing the control requirements of a turbofan engine, selecting the combustion chamber fuel F and the nozzle area A8 of the turbofan engine as manipulated variables according to the mechanism of the turbofan engine, and selecting high-speed rotor speed N2 and exit pressure ratio πT as controlled variables, i.e. U=[F, A8]T and Y=[N2,πT]T;
S2. selecting the first group of manipulated variables and controlled variables to perform analysis, maintaining the input variable A8=0.2602m2 of the turbofan engine, and setting the value of the first group of manipulated variables to
{tilde over (F)}=[{tilde over (F)}1, {tilde over (F)}2, . . . , {tilde over (F)}320]
where the monotonically increasing condition is met and the value is within a reasonable range;
S3 running a turbofan engine model to obtain output data
Ñ2=[Ñ2,1, Ñ2,2, . . . , Ñ2,320]
S4. normalizing the obtained input {tilde over (F)} and output Ñ2 of the turbofan engine as follows:
S5. using the correlation analysis method to calculate a correlation coefficient Re1 as follows:
S6. obtaining Re1=0.9916>0.5 through the above-mentioned calculation, which indicates that the correlation between two variables is high, and the two variables can be used as a group of manipulated variables; similarly, the correlation coefficient Re2=0.9979>0.5 of A8 and πT can be obtained, so the variables can also be used as the second group of manipulated variables; if the correlation coefficient between a manipulated variable and a controlled variable is less than 0.5, the manipulated variable needs to be replaced until the correlation coefficient is satisfactory.
As shown in
S1. establishing a fastest control integrated function ƒhan(p,q,rt, h) of a second-order discrete system, and the expressions thereof are as follows:
where, rt is a parameter to be tuned and is used for tuning the length of the transition time for the output variables of the TD; the greater rt is, the shorter the transient time is; h is a simulation step size; sign(ƒx) is a sign function,
d, d0, ƒx x,a0,a are internal variables introduced for the convenience of calculation, and ƒhan is the output of the function ƒhan(p,q,rt,h);
S2. building the TD module according to the established function ƒhan(p,q,rt,h), the input of the TD is a control instruction vi of a control system, the output is respectively a tracking trajectory gi and a trajectory derivative gi′, and the discrete-form updated expressions thereof are as follows:
gi=gi+h×gi′
gi′=gi′+h׃han(gi−vi, fi′, rt,i, h)
S3. establishing the LESO which is used to observe the states of the current output variables and the total disturbance according to the current input ui and output yi; the expressions thereof are shown as follows:
where, b0,i and w,i oare respectively parameters to be tuned, b0,i , is a model characterization parameter and is related to an actual model, and wo,i is a bandwidth parameter of the ESO; ui and yi are output of the ESO, Z is a state variable of the ESO, Ŷ is the estimated output of the ESO, and three output variables are respectively the estimated value ŷe,i of the output yi of the turbofan engine, the changing trend ŷe,i′ of yi, and the total disturbance ŷtotal,i of the turbofan engine in this loop;
S4. establishing a nonlinear function ƒal(e, α, δ) , and the expression thereof is shown as follows:
S5. establishing a nonlinear PD feedback controller shown as follows in accordance with the established nonlinear function:
uPD,i=Kp,iƒal(ei, αp,i, δp,i)+Kd,iƒal(ei′, αd,i, δd,i)
S6. assembling and establishing the steady state and transition state controller of the turbofan engine in accordance with the structure of each module in
As shown in
S1. preliminarily selecting the model characterization parameter b1,0=200 in the ESO according to the data {tilde over (F)} and Ñ2 of the turbofan engine to meet the condition of the following expression:
S2. preliminarily selecting wo,1=1 of the ESO, using the data {tilde over (F)} and Ñ2 of the turbofan engine as the input of the ESO, and observing whether the estimated value {circumflex over (N)}2,e of the output thereof, i.e. N2 can track Ñ2 correctly; if the tracking effect is poor, increasing the value of wo,1, and observing the tracking effect; if the tracking effect is still poor when wo,1 is relatively great (for example wo,1>100), decreasing b0,1 moderately, and observing the tracking effect after retuning until the observed effect is good;
S3. inputting a reference instruction v1, and tuning a parameter rt,1 to obtain a desired transition trajectory gi; when the trajectory transition time is too long, increasing the parameter rt,1; otherwise, decreasing rt,1;
S4. using e1=g1−ŷe,1 and e1′=g1′−ŷe,1′ as the input of a nonlinear PD controller, and tuning Kp,1 and Kd,1 to reasonable values; the output thereof is uPD,1;
S5. calculating the output of the steady state and transition state controller as
S6. comprehensively tuning Kp,1, Kd,j, wo,1 and b0,1 to ensure that the controlled variable N2 can track a given trajectory g1 well; after repeated tuning, the finally obtained parameter combination with better effect is Kp,1=600, Kd,1=30, b0,1=200, wo,1=40, rt,1=5000; similarly, the effect is better when the parameters of A8 and πT control loops are Kp,2=50, Kd,2=10,b0,2=3000, wo,2=30, rt,2=1.
As shown in
S1. determining the temperature limit T4,1C,l=1700K and the parameter value T4,1C,dl=30K when a limit protection controller comes into action according to the control objectives of the turbofan engine;
S2. establishing the upper limit protection function ƒunl(T4,1C, T4,1C,l, T4,1C,dl, Fl) according to the formula, i.e.
S3. establishing a discrete-state dead band link based on the limit protection function, and the expression thereof is shown as follows:
where, the two parameters of the dead band are respectively selected as Δl=2, Δl′=5;
S4. connecting the output of the limit protection controller to the control loop in the form of negative feedback, and selecting the maximum value Fl=300 kg/h that the limit protection controller can output to ensure that the limit protection controller can fulfill the role of over-limit protection.
The control effect of a steady state and transition state multi-variable controller of a turbofan engine based on an active disturbance rejection control theory (ADRC) after the design is completed is shown in
For the disturbance rejection control performance of this method, the turbofan engine is operated to the declared working condition without changing the controller parameters, and the influence of the disturbance on the control effect is observed by applying augmentation fuel.
For the performance test of the limit protection controller, since this controller will not be triggered under normal circumstances, the limiting value of the limit protection parameter needs to be lowered to observe the effect thereof Here, the temperature limit protection is modified to T4,1C,l=1600K, and the control effect is shown in
In conclusion, using the steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC) can meet the control requirements of the turbofan engine and ensure that the turbofan engine operates within a safe range.
Claims
1. A steady state and transition state multi-variable control method of a turbofan engine based on an active disturbance rejection control theory (ADRC), comprising the following steps: u ⋓ i, k = u ~ i, k - min ( u ~ i ) max ( u ~ i ) - min ( u ~ i ) y ⋓ i, k = y ~ i, k - min ( y ~ i ) max ( y ~ i ) - min ( y ~ i ) u _ i = 1 m ∑ k = 1 m u ⋓ i, k y _ i = 1 m ∑ k = 1 m y ⋓ i, k var ( u ⋓ i ) = ∑ k = 1 m ( u ⋓ i, k - u _ i ) 2 var ( y ⋓ i ) = ∑ k = 1 m ( y ⋓ i, k - y _ i ) 2 Re i = ∑ k = 1 m ( u ⋓ i, k - u _ i ) ( y ⋓ i, k - y _ i ) var ( u ~ i ) var ( y ~ i ) d = r t × h d 0 = h × d fx = p + hq a 0 = d 2 + 8 r t fx a = { q + a 0 - d 2 sign ( fx ), fx > d 0 q + fx h, fx ≤ d 0 f han = - { r t × sign ( a ), a > d r t a d, a ≤ d sign ( fx ) = { 1, fx ≥ 0 - 1, fx < 0, Z = [ z 1, i, z 2, i, z 3, i ] T Y = [ y ^ e, i, y ^ e, i ′, y ^ total, i ] T Z. = [ - 3 w o, i, 1, 0 - 3 w o, i 2, 0, 1 - w o, i 3, 0, 0 ] Z + [ 0, 3 w o, i b 0, i, 3 w o, i 2 0, w o, i 3 ] [ u i y i ] Y ^ = [ 1, 0, 0 0, 1, 0 0, 0, 1 ] Z fal ( e, α, δ ) = { e δ 1 - α, e ≤ δ e α sign ( e ), e > δ min ( y ~ i ) max ( u ~ i ) ≤ b 0, i ≤ max ( y ~ i ) min ( u ~ i ) u i = u PD, i - y ^ total, i b 0, i fun l ( x i, x i, l, x i, dl, u j, l ) = { 0, x i ≤ x i, l - x i, dl u j, l × x i, dl x i - x i, l + 1 x i, dl - 0.1 x i, dl + 1, x i, l - x i, dl < x i < x i, l - 0.1 x i, dl u j, l, x i ≥ x i, l - 0.1 x i, dl fun out ( k ) = { fun l ( x i, k, x i, l, x i, dl, u j, l ), if x i, k - x i, k - 1 > Δ i or x i, k + x i, k - 2 - 2 x i, k - 1 > Δ i ′ fun l ( x i, k - 1, x i, l, x i, dl, u j, l ), otherwise
- S1. selecting two or more groups of controlled variables and determining the control parameter requirements of each controlled variable based on multi-variable control objectives, and then determining a manipulated variable corresponding to each controlled variable in combination with a mechanism analysis method and a correlation analysis method;
- the steps of determining the controlled variables and the manipulated variables are as follows:
- S1.1 analyzing the control requirements of a turbofan engine, and preliminarily determining the manipulated variables U=[u1,..., ui,..., un]T and the controlled variables Y=[y1,..., yi,..., yn]T according to the mechanism of the turbofan engine, wherein ui and yi are the variables of the same group in the ith loop;
- S1.2 selecting the ith group of manipulated variables and controlled variables to perform analysis, maintaining other input variables [u1,..., ui−1, ui+1,..., un] of the turbofan engine as reasonable constants, and setting the sequence of the ith group of manipulated variables to ũi=[ũi,1, ũi,2,..., ũi,k,..., ũi,m]
- where: ũi,k−1<ũi,k, ũi,k is within the normal operation range of the manipulated variable ui, and m means the number of data sampling points in this group;
- S1.3 running a turbofan engine model to obtain output data {tilde over (y)}i=[{tilde over (y)}i,1,... {tilde over (y)}i,m]
- S1.4 normalizing the obtained input ũi and output {tilde over (y)}i of the turbofan engine as follows:
- S1.5 using the correlation analysis method to calculate a correlation coefficient Rei as follows:
- S1.6 determining whether the manipulated variable ui is used for controlling the controlled variable yi according to the obtained correlation coefficient Rei; if the correlation coefficient is satisfactory, this group of parameters is selected correctly; otherwise, if the correlation between the manipulated variable and the controlled variable is not significant, the manipulated variable needs to be replaced until the correlation coefficient is satisfactory;
- S2. establishing a tracking differentiator (TD) module, a linear extended state observation module and a nonlinear PD module according to the basic principle of active disturbance rejection control, building a steady state and transition state controller, and reserving the time constant rt of the TD, wo of the linear extended state observer (LESO), and Kp and Kd of the nonlinear PD as the parameters to be tuned;
- the steps of establishing the steady state and transition state controller are as follows:
- S2.1 establishing a fastest control integrated function ƒhan(p,q,rt,h) of a second-order discrete system, and the expressions thereof are as follows:
- where, rt is a parameter to be tuned and is used for tuning the length of the transition time for the output variables of the TD; the greater rt is, the shorter the transient time is; h is a simulation step size; sign(ƒx) is a sign function,
- d, d0, ƒx, a0, a are internal variables introduced for the convenience of calculation, and ƒhan is the output of the function ƒhan(p,q,rt,h);
- S2.2 building the TD module according to the established function ƒhan(p,q,rt,h), the input of the TD is a control instruction vi of the ith loop, the output is respectively a tracking trajectory gi and a trajectory derivative gi′, and the discrete-form updated expressions thereof are as follows: gi=gi+h×gi′ gi′=gi′+h׃han(gi−vi, gi′, rt,i, h)
- S2.3 establishing the LESO, using the input ui and output yi of the turbofan engine as the input of extended state observation, and observing the states ŷe,i and ŷe,i′ of the current output variables and the total disturbance ŷtotal,i; the expressions thereof are shown as follows:
- where, b0,i and wo,i are respectively parameters of the extended state observer (ESO), b0,i is a model characterization parameter and is related to an actual model, and wo,i is a bandwidth parameter of the ESO; ui and yi are the input of the ESO, Z is a state variable of the ESO, Ŷ is the output of the ESO, and three output variables are respectively the estimated value ŷe,i of the output yi of the turbofan engine, the changing trend ŷe,i′ of yi, and the total disturbance ŷtotal,i of the turbofan engine in this loop;
- S2.4 establishing a nonlinear function ƒal(e,α,δ), and the expression thereof is shown as follows:
- S2.5 establishing a nonlinear PD feedback controller shown as follows in accordance with the established nonlinear function: uPD,i=Kp,iƒal(ei, αp,iδp,i)+Kd,iƒal(ei′, αd,i, δd,i)
- S2.6 assembling and establishing the steady state and transition state controller of the turbofan engine in accordance with the structure of each module in the ADRC;
- S3. selecting a group of manipulated variables and controlled variables as a tuned parameter control loop, maintaining the manipulated variables of the remaining loops unchanged or operate in accordance with a desired trajectory, and tuning the parameters of the steady state and transition state controller to basically meet the control requirements;
- the steps of tuning the ith group of control parameters of the steady state and transition state controller are as follows:
- S3.1 preliminarily selecting the parameters of the ESO according to the data ũi and {tilde over (y)}i of the turbofan engine to meet the condition of the following expression:
- S3.2 using the data ũi and {tilde over (y)}i of the turbofan engine as the input of the ESO, and tuning the parameter wo,i of the ESO to make the output ŷe,i of the ESO track {tilde over (y)}i correctly;
- S3.3 inputting a reference instruction vi, and tuning a parameter rt,i to obtain a desired transition trajectory gi; when the trajectory transition time is too long, increasing the parameter rt,i; otherwise, decreasing rt,i;
- S3.4 using ei=gi−ŷe,i and ei′=gi′−ŷe,i′ as the input of a nonlinear PD controller, and tuning Kp,i and Kd,i to reasonable values; the output thereof is uPD,i;
- S3.5 calculating the output of the steady state and transition state controller as
- S3.6 forming a closed loop by the controller and the turbofan engine, and then comprehensively tuning Kp,i, Kd,i, wo,i and b0,i to ensure that the controlled variable yi can track a given trajectory well;
- S4. establishing an upper limit protection function ƒunl(xi, xi,l, xi,dl, uj,l), designing a state/performance parameter protection controller of the turbofan engine based on this function, and limiting the output of the control loop to ensure that the controlled variables of the system meet the control requirements while the state parameters thereof do not exceed the limit;
- the steps of establishing the state/performance parameter protection controller of the turbofan engine are as follows:
- S4.1 determining the limiting value xi,l of a limit protection parameter and the action range xi,dl of a limit protection controller according to the control objectives of the turbofan engine;
- S4.2 establishing the upper limit protection function ƒunl(xi, xi,l, xi,dl, uj,l),and the formula thereof is as follows:
- where, xi is a turbofan engine parameter requiring limit protection, xi,l is the allowable maximum value of xi, xi,dl is the size of the operating range of the limit protection controller, i.e. the controller comes into action when xi>xi,l−xi,dl, and uj,l means the maximum value that the limit protection controller can output, wherein j means that the controller output corresponding to the limit protection parameter is not related to the sequence of the manipulated variables of the turbofan engine;
- S4.3 establishing a discrete-state dead band link based on the limit protection function, and the expression thereof is shown as follows:
- where, ƒunout (k) means the output of the limit protection controller at the kth time point, xi,k, xi,k−1, xi,k−2 are respectively the numerical values of the turbofan engine parameter at the kth, (k−1)th and (k−2)th time points, and Δi and Δi ′ are respectively the sizes of a position dead band and a speed dead band;
- S4.4 connecting the output of the limit protection controller to the control loop in the form of negative feedback, and tuning the maximum value uj,l that the limit protection controller can output to ensure that the limit protection controller can fulfill the role of over-limit protection;
- S5. conducting a multi-variable controller test, and finely tuning each parameter to ensure the overall effect of the turbofan engine so as to enhance the adaptability of a multi-variable control system of the turbofan engine to the change of external environment.
Type: Application
Filed: Dec 11, 2018
Publication Date: Jun 11, 2020
Patent Grant number: 11392094
Inventors: Xian DU (Dalian City), Yanhua MA (Dalian City), Ximing SUN (Dalian City)
Application Number: 16/462,521